Tesla Software updates - Australia

[ZeeDoktor]

@Vostok and @RichardV you are of course correct that all app related comms goes via the Tesla VPN, I didn't think of that when I made my earlier statements. But the mystery isn't that easily solved... Just did a quick test on the browser in the car and it's connecting from 1.144.x.x range of Telstra IP's, so indeed Telstra are the local provider. That however still does not preclude that Tesla have bulk purchased from a different (overseas) provider and the cars are in fact roaming here. Many IoT companies are doing deals of that sort.

Going to speedtest.net with the built in browser is showing excessively long ping times for me (700ms and more) and the bandwidth is atrocious - but that is true both with the car connected to Wifi with my Telstra AP, and with the car's own LTE, so could be down to the embarrassingly underspecced Atom processor used for the MCU.

Have a look at how the attached IP architecture in LTE networks diagram. The mobile node in a foreign network is routing all traffic via the home network (which is why our roaming overseas is dead beat unusuable) - but will still show a local provider's IP address as the end point.

 

[cafz]

The mobile node in a foreign network is routing all traffic via the home network (which is why our roaming overseas is dead beat unusuable) - but will still show a local provider's IP address as the end point.

No, that's not how IP networks work. Telstra owned addresses are routed to Telstra's ASNs. If you're seeing connections from a Telstra address, it's going via Telstra and Telstra's upstreams.

We know that Tesla are using Telstra SIMs in Australia anyway, because there was some mention a while ago of the process by which the Telstra SIMs are shipped to Shanghai to be installed (this was in connection with that bad batch of SIMs you might remember from a couple of years ago).

 

[Vostok]

Pretty sure Tesla had to set up servers in China separately for that reason, as China was worried about Tesla cars spying on them.

I forgot about China - yes, Tesla use a different API address within China (and only within China) for that reason (https://owner-api.vn.cloud.tesla.cn). Everyone else talks to California (https://owner-api.teslamotors.com).

 

[Vostok]

Have a look at how the attached IP architecture in LTE networks diagram. The mobile node in a foreign network is routing all traffic via the home network (which is why our roaming overseas is dead beat unusuable) - but will still show a local provider's IP address as the end point.

Um, no. In a mobile network a roaming SIM will need to authenticate with their home network (since only the home network can do that), and what that diagram is showing is that the authentication traffic goes back to the home network through a VPN tunnel.

But once authenticated, user plane data traffic is not routed through the home network. That would be silly, and doesn’t happen.

 

[ZeeDoktor]

@cafz @Vostok you both speak with the certainty of experts - which is great, because I am not. Perhaps you could elaborate on your responses? I have suffered through a month long ITU LTE and 5G course a few years ago which is where my apparently incorrect knowledge comes from and I by all means don't claim certainty about any of this. I don't have any practical IMT networking experience except as a user. But plenty of practical TCP/IP background.

Can you explain to us why, if not for the reasons I posted, when roaming with an overseas provider I get a fraction of the network performance even after authentication with the home network when compared to when using the same device with a network native SIM? This was explained to us exactly as per the "that would be silly" method: all traffic goes through the home network and is forwarded to the care-of address in the roaming network.

I see the same effect when roaming overseas with my aussie mobile, as when roaming here with an overseas mobile. Performance is terrible both ways.

 

[1001]

AU-2022.36-14191 download in progress

I wonder if this is now a dedicated AU fork in the software?

 

[1001]

Bah navigation update

 

[Chuq]

Yeah, I was going to say, sounds like map data.

 

[blurrybird]

M3P on 2022.28.2 was prompted at 1PM UTC+11 to get 2022.36.5.

Installing now.

 

[Baobab]

I so hoped that when delivered, my MY will have the latest update but nooo... 2022.23.101 and after 2 days still sitting on it without any sign of updating itself

 

[Didaho]

Interesting. I wonder if this has some national security implications.

Four eyes would know what the fifth was getting up to in the back seat? They could probably just ask Siri or Alexa.

 

[Zigman]

I so hoped that when delivered, my MY will have the latest update but nooo... 2022.23.101 and after 2 days still sitting on it without any sign of updating itself

As I understand it (happy to be corrected) firmware that is pushed to end users is generally a multiple of 4 (i.e every 4th week, hence we are seeing 2022.28.xx, 2022.32,xx 2022.36.xx 2022.40.xx, etc - although I think they jumped .32) where xx is generally bug fixes/tweaks for that release. Your MY with 2022.23.xx is the release installed at the factory. I'd expect it to jump to 2022.36.xx or 2022.40.xx soon'ish* (* "soon'ish" is a very technical term when someone has no real idea of exactly when )

 

[Priit]

M3P on 2022.28.2 was prompted at 1PM UTC+11 to get 2022.36.5.

Installing now.

Oh happy day, mine is also on 2022.28.2 and appears to show that an update is available to download to 2022.36.5, hope this fixes the issues I have been having lately, so will install when I get home.

As I understand it (happy to be corrected) firmware that is pushed to end users is generally a multiple of 4 (i.e every 4th week, hence we are seeing 2022.28.xx, 2022.32,xx 2022.36.xx 2022.40.xx, etc - although I think they jumped .32) where xx is generally bug fixes/tweaks for that release. Your MY with 2022.23.xx is the release installed at the factory. I'd expect it to jump to 2022.36.xx or 2022.40.xx soon'ish* (* "soon'ish" is a very technical term when someone has no real idea of exactly when )

The first number is year, second is the week of the year and yes, they tend to do major versions every 4 weeks, though sometimes skip over one or two batches, usually includes larger changes or additional functions. The 3rd number is a revision or fix number and seems just sequentially get larger and often skips a few on major rollouts. Though sometimes the 3rd number changes also include new features. Sometimes you also get 4th number that is usually just minor bug fixes for that version

 

[Baobab]

Thank you, good gentlemen, it makes sense now.

 

[moa999]

I so hoped that when delivered, my MY will have the latest update but nooo... 2022.23.101 and after 2 days still sitting on it without any sign of updating itself

As above. Odd 2nd number denotes a factory version.
It's likely equivalent to a .28 or .36 production given the highish third number.

Sometimes the car downloads a new version at the delivery centre. Other times it can be a month or so.

 

[Maximillan]

I so hoped that when delivered, my MY will have the latest update but nooo... 2022.23.101 and after 2 days still sitting on it without any sign of updating itself

In order to ensure that they can get it out of the factory they ship with a version with known bugs.

Now that it's in your hands, welcome to the testing program!

 

[Priit]

Oh happy day, mine is also on 2022.28.2 and appears to show that an update is available to download to 2022.36.5, hope this fixes the issues I have been having lately, so will install when I get home.

So... Is it just me or have the updates got a lot longer than they used to be installation time wise? I remember previous updates take an hour or so usually, some quick ones even managed in half an hour. This took ages. It did the 50% download section as usual withing 2-3 minutes. I suspect thats when it actually just downloads the update and I have a reasonably fast internet. Then the remaining 50% would be some form of checksum verify process and unpacking of the download I assume, as this time it took 1.5 hours (that's 90 minutes) So rather longer than usual. Then as soon as that got done I tapped the update and it took another 65 minutes for the installation. so the whole process took about 2 hours and 40 minutes. I wonder if that's because my previous software version was rather glitchy on the communications side, so it may have had trouble verifying its software if it needed to call home to verify or if the updates have got way more comprehensive and just take longer to install.
I have not had a chance to take it out for a spin, but I can confirm the sentry remote viewing option was in the settings, so that is new for me and hopefully the communications system will be more stable again, as that has been lately my main issue with it.

 

[Hairyman]

I noticed that going to .36.5 but not .36.6

 

[Vostok]

I see the same effect when roaming overseas with my aussie mobile, as when roaming here with an overseas mobile. Performance is terrible both ways.

• Australian mobile network speeds are world class - we are usually in the top 10 countries globally. So when your Australia mobile device travels overseas, if you are getting rubbish performance it’s because most mobile networks overseas are just that. There’s also possibly a spectrum mismatch (see the next point).
• When an overseas mobile visits here, the device probably can’t access all the same spectrum bands we have. Foreign devices will be different model numbers that support a different set of spectrum bands used in the region it comes from, only some of which might be in use here. So the device has access to fewer spectrum bands than native devices and will therefore experience much slower speeds.

Neither of these outcomes has anything to do with how user plane traffic is routed.

 

[ZeeDoktor]

Nope, that's not the case. I'm specifically comparing Switzerland's Telstra (Swisscom, the former national telco) SIM roaming on Telstra network here, and Telstra SIM roaming on Swisscom network there. Both in the same device (iPhone 12 Pro). Both SIMs are giving a small fraction of the non-roaming network performance when roaming in each other's countries. Latency in the 100s of ms, while roaming, low teens of ms when not roaming.

Worldclass performance? Where do you get this in Oz on an iPhone with 5G? Speedtest by Ookla - The Global Broadband Speed Test